ChatGPT and Language Models: Assessing Emergent Abilities and their Potential Existential Threat to Humanity

Large Language Models, like ChatGPT, have demonstrated the ability to perform a wide variety of tasks, when trained primarily on the task of predicting masked words in sentences. This remarkable set of capabilities, known as Emergent Abilities, has enabled their widespread adoption across a range of tasks and domains.

In this talk, Dr. Tayyar Madabushi will offer a concise overview of these models, including their capabilities. He will explore how these models represent a significant departure from previous machine learning systems, and why global leaders view them as potentially posing an existential threat to humanity. Finally, the talk will describe research conducted at the University of Bath which was included in the discussion paper for the UK AI Safety Summit at Bletchley Park and has since been presented at the Turing Institute and to the US Army Research Lab. Our work explains the underlying mechanisms of these models, providing a blueprint for effective deployment across various tasks and domains.

Bio

Dr. Tayyar Madabushi's research focuses on understanding the fundamental mechanisms that underpin the performance and functioning of Large Language Models such as ChatGPT. His work was included in the discussion paper on the Capabilities and Risks of Frontier AI, which was used as one of the foundational research works for discussions at the UK AI Safety Summit held at Bletchley Park.

In addition, his work on language models includes collaborative industrial research aimed at rectifying biases in speech-to-text systems widely utilised across the UK. Before starting his PhD in automated question answering at the University of Birmingham, Dr. Tayyar Madabushi founded and headed a social media data analytics company based in Singapore.